| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Neuroradiology |
1 From the Department of Radiology, University of Bonn, Sigmund-Freud-Str 25, D-53105 Bonn, Germany. Received August 19, 2003; revision requested October 31; final revision received February 26, 2004; accepted April 6. Address correspondence to C.K.K. (e-mail: kuhl@uni-bonn.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: The study design was approved by the institutional review board, and all patients gave informed consent. In a prospective intraindividual study, 25 patients who had clinical symptoms consistent with ischemic stroke underwent DW MR imaging at both 1.5 T and 3.0 T. The 3.0- or 1.5-T examination was performed immediately one after the other, in random order. Two readers in consensus recorded the presence and number of ischemic lesions and rated image quality and lesion conspicuity. The image SNR and the CNR of the ischemic lesions were quantified. Paired Student t and Wilcoxon matched-pairs signed rank tests were used to test for statistical significance.
RESULTS: Image quality at 3.0-T DW MR imaging was consistently lower than that at 1.5-T DW MR imaging owing to greater image distortions (P < .05). Yet, overall SNR and lesion CNR at 3.0 T increased significantly; mean increases were 48.8% (P < .001) and 96.3% (P < .01), respectively. The higher overall SNR and lesion CNR translated into a significantly higher sensitivity in the detection of ischemic lesions at 3.0 T than at 1.5 T. Of the total of 48 lesions that were identified in 19 of the 25 patients, 47 (98%) were diagnosed at 3.0 T and 36 (75%) were diagnosed at 1.5 T. In addition, the conspicuity of the lesions that were visible with both systems was significantly higher at 3.0 T (P < .001).
CONCLUSION: Although 3.0-T DW MR imaging generates greater image distortions, it yields increased SNR and CNR compared with DW MR imaging at 1.5 T. The increased CNR at 3.0 T translates into a significantly improved diagnostic confidence in the detection of focal apparent diffusion coefficient changes in the setting of subacute and acute ischemic stroke.
© RSNA, 2004
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
So far, most experience with diffusion imaging is at 1.5 T. Since the U.S. Food and Drug Administration approved the use of higher magnetic field strengths for MR imaging, high-field-strength systems operating at 3.0 T and higher have become increasingly available in clinical settings. Higher magnetic fields promise to yield at least linear increases in SNRs and probably contrast-to-noise ratios (CNRs), yet they also cause specific difficulties: For example, magnetic susceptibility scales exponentially as the field strength increases, and, accordingly, image distortions may become severe. With echo-planar MR imaging pulse sequences, which are typically used in diffusion imaging, these artifacts would deteriorate image quality, especially in areas close to the skull base.
The objective of this study was to compare SNR, CNR, image quality, and confidence in diagnosis for diffusion imaging at 1.5 T and 3.0 T.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patients and Inclusion Criteria
Patients were eligible for entry into this study if they (a) had clinical symptoms consistent with acute or subacute ischemic stroke, (b) had no evidence of cerebral hemorrhage at recently performed computed tomography, (c) were not eligible for intraarterial thrombolytic therapy, (d) had no metallic implants (eg, aneurysm clips, vascular clips, and pacemakers), and (e) were able to communicate and provide informed consent. A total of 26 consecutive patients met all inclusion criteria and were therefore considered eligible for study entry.
One patient, a 57-year-old man, withdrew his consent and refused to undergo the second MR imaging examination, which in his case was the 1.5-T examination. Thus, a total 25 patients, 16 men and nine women, were available for assessment. The male patients had a mean age of 54 years (median, 58 years; range, 39–82 years). The female patients had a mean age of 62 years (median, 70 years; range, 37–82 years). The entire cohort of 25 patients had a mean age of 57 years (median, 60 years; range, 37–82 years). The age distribution was not significantly different (t test for independent variables) between the male and female study participants. At DW MR imaging, 19 of the 25 patients eventually received a diagnosis of focal ADC changes consistent with ischemia. The 19 patients had a total of 48 such lesions.
In the 25 patients, the mean interval after the onset of ischemic symptoms and the DW MR imaging examinations was 82 hours (range, 11 hours to 7 days). According to current clinical definitions, seven of the 25 patients received a diagnosis of acute ischemia (time interval between onset of clinical symptoms and DW MR imaging, <48 hours) and 18 received a diagnosis of subacute stroke (time interval between onset of clinical symptoms and DW MR imaging, 48 hours to 10 days) at clinical examination. The patients were referred from our institution’s stroke service and had symptoms of acute or subacute ischemic stroke, including a variety of motor and sensory neurologic deficits. The identified risk factors (several per patient) for ischemic stroke were as follows: known arteriosclerosis of brain-supplying arteries in 19 patients; arterial hypertension in 17 patients; hypercholesterolemia in 12 patients; atrial fibrillation in three patients; patent foramen ovale in three patients; and hyperhomocystinemia, vasculitis, and dissection of brain-supplying arteries in one patient each.
Imaging
All 25 patients in the study cohort underwent DW MR imaging twice—once at 1.5 T and once at 3.0 T. To avoid the possibility of any time-dependent changes in ADC interfering with the study results, the 3.0-T examination and the 1.5-T examination were performed immediately one after the other in a randomized order: 12 patients were randomly assigned to undergo 3.0-T imaging first, and the remaining 13 patients were randomly assigned to undergo 1.5-T imaging first. The average time interval between the start of the first and the start of the second DW imaging examination was 32 minutes (range, 4–242 minutes).
The imaging examinations were performed by using 1.5- and 3.0-T whole-body MR imaging systems (Intera 1.5 T and Intera 3.0 T, respectively; Philips Medical Systems, Best, the Netherlands). Both systems were equipped with high-performance gradients with a maximum slew rate of 150 mT/m/msec and a maximum strength of 30 mT/m. A standard transmit-receive birdcage head coil was used for the 3.0-T examinations, and a standard receive-only birdcage coil was used for the 1.5-T examinations.
DW imaging was performed with both systems by using a single-shot spin-echo echo-planar MR imaging pulse sequence and the following parameters: 4345/82 (repetition time msec/echo time msec), transverse plane, a 128 x 128 matrix, 20 sections, a 5-mm section thickness, and b values of 0, 600, and 1000 sec/mm2. With all three b values, the diffusion sensitization was repeated in each orthogonal gradient direction (ie, phase-encoding, readout, and section-select directions) and the final isotropic diffusion maps were automatically calculated by the system, which averaged the three measurements. In addition, ADC maps were calculated for each examination.
The diffusion examinations were integrated into the regular clinical stroke imaging protocol, which was tailored to and thus varied somewhat with each clinical situation. However, with both the 1.5-T and the 3.0-T magnets, all patients were examined at T2-weighted turbo spin-echo (3540/100, 31 sections, 512 x 400 matrix at 3.0 T, 256 x 256 matrix at 1.5 T), fluid-attenuated inversion-recovery (12 000/140/2800 [repetition time msec/echo time msec/inversion time msec] at 3.0 T and 6000/100/2000 at 1.5 T, 256 x 256 matrix, coronal and transverse planes), and T1-weighted spin-echo (500/15, 25 sections, 256 x 256 matrix) MR imaging before and after the injection of 0.1 mmol of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) per kilogram of body weight.
Data Validation
The diagnoses were validated by using clinical and MR imaging follow-up and by using the patients’ ADC map data, the latter of which proved to be necessary because of the known reversibility of lesion hyperintensity on DW MR images obtained in patients with transitory neurologic deficits. A hyperintense lesion seen on a DW MR image was considered to represent a true ischemic lesion if on the corresponding ADC map a signal intensity decrease was observable. In addition, during region-of-interest (ROI) analysis (described in next paragraph), we also copied the ROI of the lesion to the corresponding ADC map to measure the ADC at the site of the presumed ischemic lesion and thus confirm a reduced ADC.
Data Analysis
Quantitative analysis: SNR and CNR.—DW MR images were downloaded to a workstation (EasyVision; Philips Medical Systems). The 1.5- and 3.0-T images obtained in the same patient were displayed on the screen. The isotropic DW images obtained at a b of 1000 sec/mm2 were used for further analysis. One of the authors (S.G.) manually drew ROIs on both images and placed them in (a) the brainstem (pons), (b) the cerebellar hemisphere, (c) the deep gray matter (thalamus), (d) the deep white matter (centrum semiovale), and (e) the ghost artifact–free part of the background noise. ROI sizes varied according to their location: The infratentorial and thalamic ROIs had an average size of 140 mm2, the centrum semiovale ROIs had an average size of 862 mm2, and the background noise ROIs had an average size of 4300 mm2.
In the 19 patients who had ischemic lesions, ROIs were drawn in the ischemic lesion and in the adjacent normal-appearing brain tissue. In all instances, care was taken to exactly match the positions of the ROIs on the two DW MR images obtained in the same patient. This process was performed by S.G. with the close supervision of a neuroradiologist, C.K.K. or J.T., each of whom had about 10 years experience reading brain MR images. The neuroradiologist marked the ischemic lesions on the film images and checked the positions of the ROIs.
For all anatomic regions, the SNR was calculated at all DW MR imaging examinations according to the following equation: SNR = SItissue/SDnoise, where SItissue is the ROI-based signal intensity of the brain tissue and SDnoise is the standard deviation of the signal intensity of the background noise. We calculated the CNR to quantify the contrast between the ischemic lesion and the adjacent normal-appearing brain tissue. The following equation was used to calculate lesion CNR: CNR = (SIlesion – SInabt)/SDnoise, where SIlesion is the signal intensity of the ischemic lesion and SInabt is the signal intensity of the adjacent normal-appearing brain tissue. The tissue SNR and lesion CNR obtained at 1.5 T were compared with the tissue SNR and lesion CNR obtained at 3.0 T in the same patient.
Qualitative analysis: image quality.—Film hard-copy isotropic DW MR images obtained at a b of 1000 mm–2 · sec and corresponding film hard-copy ADC maps were acquired with standardized window settings and randomly presented to the two experienced neuroradiologists (J.T., C.K.K.). The readers were blinded to the field strength used and to the identity of the patients. The readers were asked to rate, in consensus and by using a five-point scale, the image quality in terms of the apparent SNR and the degree of image distortion due to susceptibility artifacts. A score of 1 (poor) was assigned in cases of nondiagnostic image quality, which was usually due to image distortions secondary to susceptibility artifacts and/or a very low apparent SNR. A score of 2 (marginal) or 3 (satisfactory) was assigned when the image had major or minor artifacts that did not interfere with the diagnosis and a low (score of 2) or adequate (score of 3) apparent SNR. A score of 4 (good) was assigned to images with no artifacts and an adequate apparent SNR. A score of 5 (excellent) was assigned when no artifacts were seen on the image and the apparent SNR was high. Mean image quality scores were compared between the 1.5- and 3.0-T DW image sets.
The two readers were asked to note, in consensus, the presence and number of ischemic lesions and to score the conspicuity of these lesions as follows: A score of 1 meant definitely no ischemic lesion was present; a score of 2, probably no ischemic lesion was present; a score of 3, equivocal; a score of 4, probably an ischemic lesion was present; and a score of 5, definitely an ischemic lesion was present. If multiple scattered or noncontiguous lesions were seen in the same territory of a brain-supplying artery, the lesion was considered a single lesion. A score of 3, 4, or 5 was used to designate an image that was positive for an ischemic lesion; a score of 1 or 2 was considered to indicate the presence of negative findings. To avoid any bias, the film hard-copy readings were divided into two sessions performed 6 weeks apart. The 3.0- and 1.5-T images obtained in the same patients were presented during the two separate reading sessions in a crossover fashion: During the first reading session, half of the cases—those of 12 of the 25 patients—consisted of images obtained at 1.5 T, and the other half—those of the remaining 13 patients—consisted of images obtained at 3.0 T. During the second reading session, the respective patients’ corresponding 3.0- or 1.5-T images were presented.
The number of ischemic lesions prospectively diagnosed by using the 1.5-T DW images (and the corresponding ADC maps) was compared with the number of ischemic lesions prospectively and independently identified on the same patients’ 3.0-T DW images. The degree of lesion conspicuity was compared between the 1.5- and 3.0-T images that were considered positive for ischemic lesions.
Statistical Analyses
Statistical analyses were performed with the two-sided paired Student t test and the Wilcoxon matched-pairs signed rank test by using computer software (SPSS, version 10.0.7; SPSS, Chicago, Ill). The paired Student t test was used to test the results regarding the numbers of ischemic lesions detected with the 1.5-T versus the 3.0-T system and to compare the SNR and the lesion CNR between the 1.5- and 3.0-T images. The Wilcoxon matched-pairs signed rank test was used to test the results regarding the image quality and lesion conspicuity scores assigned to images obtained at 1.5 T versus at 3.0 T. P < .05 indicated a significant difference. To account for the clustering of data in patients with multiple lesions (11 of the 19 patients had more than one lesion), a mean lesion CNR and a mean lesion conspicuity score were calculated for each patient, and this mean value was used for further analysis.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In six of the 25 patients, no abnormality was identified on either the 1.5-T or the 3.0-T images; these patients were considered to have experienced a transitory ischemic attack without structural brain damage secondary to ischemia. In accordance with this diagnosis, all six patients had a complete alleviation of their symptoms. Of interest, five patients who, on the basis of their reversible symptoms, received a clinical diagnosis of transient ischemic attack had definitive ischemic foci on their DW MR images consistent with cerebral infarction with transient signs (according to criteria of Waxman and Toole [6]).
ROI-based Quantitative Image Analysis
SNR in normal cerebral tissue and ischemic lesions.—As shown in Figure 1, all cerebral tissues showed a consistent substantial increase in signal intensity at 3.0-T DW MR imaging compared with the signal intensity seen at 1.5-T DW MR imaging. The mean SNR increase at 3.0 T, as compared with the reference SNR at 1.5 T, was 48.8% (51% in the cerebellum, 61% in the brainstem, 71% in the thalamus, 30% in the striatum, and 31% in the deep white matter)—that is, there was an almost 1.5-fold increase in SNR at 3.0 T as compared with the SNR at 1.5 T. The mean SNR in lesions at 1.5 T was 49.0 ± 12.6 (standard deviation), as compared with a mean SNR of 83.5 ± 23.5 at 3.0 T. Accordingly, the SNR in the ischemic lesions increased by an average of 80.7% (median, 82.4%) when the field strength was changed from 1.5 to 3.0 T. The difference between the SNR obtained at 3.0 T compared with that obtained at 1.5 T in the cerebral tissues, as well as in the ischemic lesion, proved to be statistically significant (P < .001, Student t test for paired samples).
|
|
|
|
|
The 35 lesions that were visible with both systems received consistently higher conspicuity scores at 3.0 T than at 1.5 T; in other words, the readers were more confident in diagnosing an ischemic lesion at 3.0 T. At patient-based analysis (ie, with use of mean conspicuity scores for the patients with multiple lesions), the mean lesion conspicuity score was 3.6 ± 1.0 at 1.5 T versus 4.6 ± 0.5 at 3.0 T (median scores, 3.75 at 1.5 T and 5.00 at 3.0 T). The difference in lesion conspicuity proved to be significant (P < .001, Wilcoxon matched-pairs signed rank test).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MR imaging in general and DW MR imaging in particular have been shown to be the most sensitive tools for the detection of ischemic stroke and even transient ischemic attacks (9–11), with a reported sensitivity for stroke detection of approximately 100% in early investigations (4,5,12,13). The ischemia-induced lack of energy-rich phosphates causes an "ischemic cascade," which by way of the breakdown of the cellular membrane potential gives rise to the cytotoxic edema that has long been known to be associated with tissue infarction. The edema, in turn, is thought to restrict diffusion and partly explains the high signal intensity at the site of ischemic tissue damage at DW MR imaging (10).
The high reliability of DW imaging for diagnosing ischemic stroke is beginning to prompt a paradigm shift in the treatment of patients (14): Guidelines for the differential therapy of ischemia have recently been reassessed to move from a merely time-based to a more tissue-based decision-making process (15,16).
However, there is increasing evidence suggesting that strokes do occur in patients with DW imaging findings that are negative for infarction (17–20), particularly during the hyperacute phase (ie, within the first 6 hours after the onset of symptoms). One possible explanation for this is the fact that DW MR images have a low SNR and that subtle ADC changes, which may be associated with early infarction, may not be detectable at DW MR imaging. This is one reason why perfusion imaging performed with T2*-weighted first-pass bolus chase or arterial spin-labeling techniques is considered an integral part of the work-up of patients who are clinically suspected of having ischemia and is still indicated in patients in whom no abnormality is identified at DW imaging (18,21).
With the increasingly important role of MR imaging for clinical decision making during the triage of patients suspected of having a hyperacute or acute stroke, there is an increasing need for the accurate ad hoc diagnosis of even subtle cerebral ischemic lesions.
Today the availability of MR imaging systems that operate at higher field strengths (>2.0 T) is consistently increasing—not only in neuroscientific research laboratories but also in clinical settings. Higher magnetic field strengths promise to yield increased SNRs and thus should be ideally suited to compensate for the inherently low SNR at DW imaging. On the other hand, whereas the SNR increases only linearly with increasing field strength, susceptibility effects are known to increase exponentially. Therefore, MR imaging pulse sequences without refocusing radiofrequency pulses will go along with substantial image distortions secondary to field inhomogeneities. This phenomenon will be even more pronounced at echo-planar pulse sequences in which the entire k-space matrix is filled with one long echo train, which will cause phase errors to accumulate, without correction, until the end of the data acquisition. DW imaging is typically performed with these types of single-shot echo-planar sequences.
Accordingly, the questions that we sought to answer in this study were as follows: (a) Is 3.0-T DW MR imaging feasible in patients who are clinically suspected of having an ischemic stroke? In particular, do the increased susceptibility artifacts, which can be expected to degrade image quality, override the diagnostic advantages that are possible owing to the higher SNR? (b) If DW MR imaging proves to be feasible at 3.0 T, is it worthwhile? Are there any clinically relevant advantages to using it rather than 1.5-T DW MR imaging?
Our study results show, in accordance with previously published material on healthy volunteers (22), that DW MR imaging at 3.0 T yields substantial susceptibility artifact–induced geometric distortions. The image quality scores of the 3.0-T DW images were significantly lower than those of the 1.5-T DW images obtained in the same patients. As predicted, the distortions were most severe in areas close to the skull base (posterior fossa and brainstem) and in the frontopolar region—that is, close to the tissue-tissue and air-tissue interfaces. The prominent image distortions in these areas were the reason why, despite the higher apparent SNR, the overall image quality at 3.0 T was consistently rated at least one score lower than that at 1.5 T in the same patient.
It is noteworthy, however, that in no patient were the distortions so severe that we believed image interpretation was impossible. Most of the 3.0-T images were of sufficient (or higher) diagnostic quality: Both the mean and the median image quality scores were 3 (satisfactory). Still, in one patient, a small infarct in the posterior fossa was visible on the 1.5-T image but not on the 3.0-T image, probably because of image distortions at high-field-strength DW MR imaging.
Thus, it seems crucial to reduce the high field strength–induced geometric distortions. There are several possible ways to achieve this: First, segmented—as opposed to single-shot—echo-planar MR imaging (23,24) can be used to avoid the accumulation of phase errors. However, performing segmented echo-planar MR imaging requires substantially longer acquisition times, which, in turn, lead to increased motion and pulsation artifacts. Navigator correction can be used in turn to reduce pulsation artifacts at segmented echo-planar MR imaging, but it also leads to unduly prolonged acquisition times, which are unacceptable, particularly for patients with acute stroke (24). The same holds true for gradient spin-echo and turbo spin-echo DW MR imaging, particularly if gated acquisitions or cardiac triggering is needed.
As an alternative, parallel imaging techniques such as sensitivity encoding promise to reduce distortions by reducing the number of phase-encoding steps necessary for image generation (25). As opposed to the use of segmented echo-planar MR imaging, the use of echo-planar MR imaging with sensitivity encoding does not lead to longer acquisition times, but it will in fact reduce them. At the time of data collection in the current study, sensitivity encoding was not available for 3.0-T DW imaging, but we anticipate that parallel imaging will be helpful in further reducing the geometric distortions and thus in preventing diagnostic "misses." The same may hold true for techniques by which phase inconsistencies at multishot fast spin-echo DW MR imaging are corrected before the multishot data are compiled to form an image (26,27).
Regarding the SNR in normal tissue, both visual image assessment and quantitative ROI-based analysis revealed a substantially higher SNR at 3.0-T DW MR imaging; measured SNR increases, from the value at 1.5 T, ranged from 31% in the deep white matter to 71% in the thalamus. These results are in good agreement with the data for healthy volunteers reported by Hunsche et al (22), who reported a mean SNR increase of 40%.
More important, however, is the finding that the SNR in ischemic lesions increased significantly, by a median of 82.4% (1.8 times higher), at 3.0-T DW imaging. The increase in SNR at 3.0-T DW imaging compared with the SNR at 1.5-T DW imaging was therefore more pronounced in the diseased (ischemic) tissue than in the healthy tissue. The lesion CNR increased even more, by an average of 96.3% and up to 238.8%. Thus, the mean CNR in ischemic lesions at 3.0-T DW imaging was almost twice as high as the mean CNR in the same lesions at 1.5-T DW imaging.
The findings in this study prompt the following question: Are the increased SNR and CNR achieved at 3.0-T DW MR imaging clinically relevant? Our results were as follows: Of the 25 patients who participated in this study, 19 received a final diagnosis of acute or subacute ischemic stroke on the basis of clinical follow-up and DW MR imaging and corresponding ADC map findings. At blinded readings of the film images obtained in as many as eight (42%) of the 19 patients, additional ischemic lesions that were not detected on the corresponding 1.5-T DW images were identified on the 3.0-T DW images. At lesion-by-lesion analysis, 47 (98%) of the total of 48 ischemic lesions confirmed to be present were prospectively diagnosed at 3.0 T and 36 (75%) were prospectively diagnosed at 1.5 T. These results suggest that 3.0-T DW MR imaging has considerably increased sensitivity for the diagnosis of subacute or acute ischemic lesions.
The additional lesions identified at 3.0 T were, without exception, small (<3–4 mm) and were much smaller than the section thickness used to perform DW MR imaging at both 1.5 and 3.0 T. We speculate that these lesions were not seen on the corresponding 1.5-T images because of the lower SNR and CNR, which, together with partial volume averaging of the adjacent normal brain tissue, precluded the prospective diagnosis on 1.5-T DW images.
All of the additional ischemic lesions that were diagnosed at 3.0 T were identified in patients who had other ischemic lesions that were also detected at 1.5-T DW imaging. Accordingly, in this subset of patients, the final diagnosis did not change owing to the additional lesions seen at 3.0 T; at patient-based analysis, 1.5- and 3.0-T DW imaging had equivalent diagnostic sensitivities: Nineteen of the 25 patients received a diagnosis of ischemic stroke with both examinations. Thus, on the basis of the current study results, it seems that 3.0-T DW imaging improves the sensitivity for the detection of small ischemic lesions. However, this higher sensitivity probably is not always needed to establish the diagnosis because most of the patients who present with clinical symptoms of stroke have ischemic lesions that are large enough to be visualized also at 1.5-T DW imaging.
It remains to be seen (and it is the objective of an ongoing larger comparative study, including more patients without clinical symptoms of stroke) whether the 3.0-T technique will enable us to identify ischemic lesions in patients who, after 1.5-T diffusion imaging, receive a diagnosis of having no infarction. Possibly, 3.0-T DW imaging may then improve our ability to identify clinically silent microembolic stroke, and, thus, may help with clinical decision making regarding the use of invasive preventive therapies (eg, carotid stent placement).
Also, it should be well understood that the majority of the patients in the present study were in the subacute stage of stroke. For ethical reasons, we could not include most of the patients who had had an acute stroke and thus may have been suitable for revascularization therapy. To our knowledge, all of the data on false-negative 1.5-T DW imaging findings have been made so far in patients with acute ischemic injury. It is conceivable that although the high sensitivity of 3.0-T DW imaging may not be needed to diagnose subacute stroke, it may be helpful to further improve the detection of acute ischemic injury. As in this period—that is, in the hyperacute setting—accurate imaging information is crucially needed to select the appropriate therapy, any imaging technique that offers a higher sensitivity for ischemia should improve our ability to exactly tailor our therapeutic approach to the individual patient's disease stage, which, in turn, should translate into better outcome.
This study had some important limitations. First and most important, it is very difficult to establish an accurate reference standard—and thus an accurate validation—of findings at DW imaging. Although MR imaging follow-up with diffusion and structural imaging was obtained in all patients and the site of the ischemic lesion was correlated with (a) the clinical picture of the patients (to determine whether the focal neurologic deficit is consistent with the lesion depicted at DW imaging) and with (b) the ADC map of the respective DW MR image, these analyses still may not have been sufficient to rule out false-positive or false-negative DW imaging findings. This is a difficulty inherent to all DW MR imaging examinations, and it should be well understood that errors due to a lack of validation cannot be excluded.
Second, we had to use two different coils: a transmit-receive birdcage coil at 3.0 T and a receive-only birdcage coil at 1.5 T. This might have introduced bias due to the more effective radiofrequency deposition in the transmit-receive birdcage coil, which may be associated with a higher SNR. It is not probable that this effect accounted for the doubled and sometimes tripled SNR that we observed with 3.0-T DW imaging, as compared with the SNR observed at 1.5 T; however, this assertion is difficult to quantify. We propose, however, that about 10%–20% of the SNR increase at 3.0 T was probably attributable to our use of the transmit-receive coil for DW MR imaging at 3.0 T.
Last, one may argue that we introduced a preselection bias by including only patients who were able to communicate and provide informed consent. This means that the patient cohort may have skewed the study sample toward less affected disease states, and, thus, the patients may have been less likely to have larger infarcts that are more readily detected at 1.5 T. Less ill patients are less likely to move during the examination, which may cause motion-induced susceptibility artifacts to be less pronounced than they would have been in a nonselected patient group. However, the patients were recruited consecutively as they were referred from our hospital’s stroke service; during the short period of patient accrual, we happened to have recruited no patient who was unable to communicate.
Still, it is possible that the patient population was skewed toward conditions that favor 3.0-T imaging in that relatively small DW image lesions (without devastating clinical sequelae) had to be identified in patients who were cooperative enough to avoid motion and thus motion-induced susceptibility effects. It is possible that in a group with more severely ill patients, patient restlessness may cause more pronounced DW image artifacts, thus possibly cancelling out some of the advantages 3.0-T DW imaging offers.
We propose two conclusions on the basis of the findings in this study:
1. High-field-strength (3.0-T) DW imaging is feasible in patients clinically suspected of having an acute or subacute stroke. The image distortions at 3.0 T are substantial compared with those at 1.5 T and may interfere with the diagnosis, particularly if there are lesions in the posterior fossa. Until advanced acquisition strategies to reduce geometric distortions, such as parallel imaging, are available, we suggest performing 3.0-T DW MR imaging in two planes to avoid false-negative diagnoses in patients who are clinically suspected of having brainstem or posterior fossa ischemia.
2. The increases in SNR and CNR at 3.0 T are significant, particularly in ischemic lesions; this means that 3.0-T DW MR imaging facilitates markedly improved detection of small ischemic lesions, with sensitivity increasing from 75% (36 of 48 lesions) at 1.5 T to 98% (47 of 48 lesions) at 3.0 T in the present study. Although the additional lesions detected at 3.0 T were in patients who had other ischemic lesions that had already been identified at 1.5 T, we believe such increased sensitivity could be clinically relevant in some cases. Further studies with other groups of patients, such as those suspected of having had an acute infarction or those without any clinical signs but with typical risk factors, are needed to investigate whether the higher sensitivity achieved at 3.0-T DW MR imaging can lead to changes in patient treatment.
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, C.K.K., H.H.S., H.U.; study concepts, C.K.K., H.H.S., J.T.; study design, C.K.K., J.T., J.G.; literature research, C.K.K.; clinical studies, C.K.K., J.T., M.v.F.; data acquisition, C.K.K., J.T., J.G., M.v.F.; data analysis/interpretation, C.K.K., J.G., J.T., S.G., M.v.F.; statistical analysis, C.K.K.; manuscript preparation and revision/review, C.K.K., H.H.S., J.G., J.T.; manuscript definition of intellectual content and final version approval, C.K.K., H.H.S.; manuscript editing, C.K.K.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. K. Kuhl, F. Traber, and H. H. Schild Whole-Body High-Field-Strength (3.0-T) MR Imaging in Clinical Practice * Part I. Technical Considerations and Clinical Applications Radiology, March 1, 2008; 246(3): 675 - 696. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Bertrand, C. Oppenheim, C. Lamy, S. Rodrigo, O. Naggara, J.L. Mas, and J.F. Meder Comparison of Optimized and Standard Diffusion-Weighted Imaging at 1.5T for the Detection of Acute Lesions in Patients with Transient Ischemic Attack AJNR Am. J. Neuroradiol., February 1, 2008; 29(2): 363 - 365. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. M. von Falkenhausen, G. Lutterbey, N. Morakkabati-Spitz, O. Walter, J. Gieseke, R. Blomer, W. A. Willinek, H. H. Schild, and C. K. Kuhl High-Field-Strength MR Imaging of the Liver at 3.0 T: Intraindividual Comparative Study with MR Imaging at 1.5 T Radiology, October 1, 2006; 241(1): 156 - 166. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. K. Kuhl, P. Jost, N. Morakkabati, O. Zivanovic, H. H. Schild, and J. Gieseke Contrast-enhanced MR Imaging of the Breast at 3.0 and 1.5 T in the Same Patients: Initial Experience Radiology, June 1, 2006; 239(3): 666 - 676. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
